Collaborative memory management for reactive sensor/actor systems

Increasing complexity of today's WSAN applications imposes demanding challenges on the underlying system design. This especially affects real-time operation, resource sharing and memory usage. Using preemptive task systems is one way to retain acceptable reactivity within highly dynamic environments. Yet, since memory is commonly rare and can often not be assigned statically, this rapidly leads to severe memory management problems among tasks with interfering and even varying requirements. Finding an allocator which suitably adapts to changing conditions while covering both issues is generally hard. We present the novel CoMem approach for maintaining high reactivity and efficient memory usage within such systems. With respect to task priorities and the typically limited performance and resources of sensor nodes, our technique facilitates compositional software design by providing tasks with runtime information for yet collaborative and reflective memory sharing. Thereby, we require no special hardware-support like MMUs but operate entirely software-based. An evaluation will show that our approach can still allow allocation delays to be close to the best case and inversely proportional to the requester's priority.